1,720,961 research outputs found
Patterns that matter
No full textPattern mining is one of the best-known concepts in Data Mining. A big problem in pattern mining is that humongous amounts of patterns can be mined even from small datasets. This makes it hard for domain experts to discover knowledge using pattern mining, for example in the field of Bioinformatics. In this thesis we address the pattern explosion using compression. We argue that the best pattern set is that set of patterns that compresses the data best. Based on an analysis from MDL (Minimum Description Length) perspective, we introduce a heuristic algorithm, called Krimp, which finds the best set of patterns. High compression ratios and good classification scores confirm that Krimp selects patterns that are very characteristic for the data. After this, we proceed with a series of well-known problems in Knowledge Discovery, which we each unravel with our compression approach. We propose a database dissimilarity measure and show how compression can be used to characterise differences between databases. We present an algorithm that generates synthetic data that is virtually indiscernible from the original data, but can also be used to preserve privacy. Changes in data streams are detected by using a Krimp compressor to check whether the data distribution has been changed or not. Finally, compression is used to identify the components of a database and to find interesting groups in a database. In each chapter, we provide an extensive experimental evaluation to show that the proposed methods perform well on a large variety of datasets. In the end, we conclude that having less, but more characteristic patterns is key to successful Knowledge Discovery and that compression is very useful in this respect. Not as goal in itself, but as means to an end: compression picks the patterns that matter
Characteristic relational patterns
No full textNowadays, relational databases have become the de facto standard to store large quantities of data. As the manual analysis of these large quantities of data is practically impossible, the field of data mining provides methods that attempts to automatically acquire insight into the data. One cornerstone technique is that of pattern mining: finding interesting regularity in data. Despite all good e orts, one can conclude that pattern mining still has a major Achilles' heel, that is, the ease at which patterns can be found. Many found patterns are slight variations on the same underlying theme, although many of them are still designated as interesting. In practice, a user gets swamped by too many similar patterns that do not contribute to a new insight into the database. In this thesis we therefore propose a di erent approach. In contrast to selecting patterns on an individual basis, we propose the selection of pattern sets. In particular, we focus on a selection scheme based on a compression technique called the Minimum Description Length (MDL) principle. The selected pattern set, our model of the data, is used to compress the complete database. According to the MDL principle, the model that compresses the database best is also the one that describes it best. As acquiring the optimal model of a database is simply too complex, we utilise a practical and heuristic approach, named Krimp. Based on this, we designed a toolbox of algorithms that derives models for di erent interpretations of the data. We discuss structured data types such as sequences and trees, the join of the database, and relational databases as a whole. These last models also show to result in good classifiers. We back up the claims in this thesis by experimental evaluation. For many of the used databases, the number of patterns initially is huge. However, we show that from this huge collection of patterns, we select a compact and good set of characteristic relational patterns
Patterns, Models, and Queries
No full textData mining provides methods that help to acquire insight in a data set automatically. One of its problem areas is to select a small set of useful patterns from the huge collection of patterns that can be found in a data set. This thesis presents our results in this area. We show that such a small set of patterns, if well-chosen, allows one to answer queries on the data set without referring to the data itself. Moreover, we show how these pattern sets allow one to built quick and scalable recommender systems. To choose such a small set of patterns, we rely on the Minimum Description Length (MDL) principle: the best model compresses the data best. More precisely, we use the code tables that the heuristic Krimp algorithm induces from the data. Our results show that these code tables are highly characteristic of the data set. Anything one wants to know about the data can be inferred from its code table. In more detail, we show how such a code table can be used to compute the answer to a query on the data set. These answers are almost always very close to the answer one gets by actually computing the query on the data itself. This similarity is verified experimentally and quantified using an asymmetric dissimilarity score which is derived from the Normalised Compression Distance. Next we show how the code tables can be used for the -- predictive -- task of tag recommendation. In particular it is shown that the proposed algorithms show a good trade-off between accuracy and time-efficiency; using the full set of patterns yields only slightly better results but requires infeasible amounts of time. In a social networking context we show how to personalize -- and thus improve -- our tag recommendations. This is achieved by using user-centred knowledge in contrast to the collective knowledge used for the general task. For quality and scalability reasons, we use `social batched personomies' by processing queries in batches, instead of individually, such as done in the standard personomy approach. In each chapter we provide extensive experimental evaluation to show that our methods perform well on a large variety of datasets. From these experiments one cannot but conclude that code tables are highly characteristic of the data
Making Pattern Mining Useful
No full textThe discovery of patterns plays an important role in data mining. A pattern can be any type of regularity displayed in that data, such as, e.g. which items are typically sold together, which genes are mostly active for patients of a certain disease, etc, etc. Generally speaking, finding a pattern is easy. Discovering interesting patterns, however, is complicated. Existing techniques for mining patterns from data lead to enormous amounts of results. Often more patterns are discovered than there is original data. As such, the problem is that it is impossible to consider the potentially interesting patterns. This thesis is about finding interesting patterns, and, more boldly, about making pattern mining useful. It is about how to discover few, but highly interesting patterns. And, prominently, it is about how to put these patterns to good use, solving a number of data mining problems. This thesis proposes to use the Minimum Description Length principle to select small groups of patterns that together describe the data well. To this end, it introduces KRIMP: a heuristic for finding the itemsets that together optimally compress the database. Through extensive evaluation, we show the high quality of these code tables. Experiments regarding the choices in the algorithm identified the best settings for the algorithm, making it parameter-free for all practical purposes. We successfully apply these patterns in a number of different data mining problems. We show how the difference between transaction databases can be measured and characterised through the use of code tables. Next, we specify the problem of identifying the parts of a transaction database drawn from different distributions in terms of MDL: the best decomposition minimises the total compressed size. We also discuss using the pattern sets as generative models. The resulting generated data preserves privacy but contains the patterns present in the original data, including correct margins. Further, we consider the problem of high quality imputation of incomplete records. Experiments show these approaches to be highly successful. By using entropy instead of frequency, the LESS algorithm we introduce is particularly suited for mining dense data. Further, by regarding data 0/1 symmetric, all major interactions in the data are captured, not just co-occurences. Experiments show that this approach is able to succinctly describe data in only tens of patterns, allowing for very easy consideration by experts. The PACK algorithm we introduce also considers data symmetrically and selects interesting itemsets. It achieves very high compression ratios. Further, it is able to mine its models directly on the data, without the need of first mining large collections of candidate patterns. The thesis concludes that to the end of making pattern mining useful simply the best set of patterns should be mined, as opposed to all patterns that satisfy certain criteria. The MDL principle is particularly well-suited for mining these useful patterns. By using this principle to select the set of patterns that describe the data best, we are returned very few, but high-quality, patterns. These characteristic patterns are easily interpreted and used in further analysis
The Computational Complexity of Probabilistic Networks
No full textIn this thesis, the computational complexity of a number of problems related to probabilistic networks is studied that combine probabilistic inference, finding, verifying, and enumerating solutions. In particular parameter tuning, sensitivity analysis, monotonicity, enumerating solutions, and problems related to qualitative abstractions of probabilistic networks are studied. These problems are not ‘merely’ NP-hard, but are complete for a variety of complexity classes in the Counting Hierarchy (CH). It is shown that these problems often remain hard under a number of constraints on the problem structure, e.g., when the treewidth of the network is bounded. This suggests, that practical applications must restrict themselves to limited degrees of freedom (e.g. a restricted number of parameters to tune or variables to determine monotonicity constraints on) in order to be tractable. Some of the problems are complete for complexity classes that have no other ‘real world’ complete problems and may be interested also from a complexity-theoretical point of view
Wiki-MetaSemantik: A Wikipedia-derived Query Expansion Approach based on Network Properties
Full text linkThis paper discusses the use of Wikipedia for building semantic ontologies to do Query Expansion (QE) in order to improve the search results of search engines. In this technique, selecting related Wikipedia concepts becomes important. We propose the use of network properties (degree, closeness, and pageRank) to build an ontology graph of user query concepts which is derived directly from Wikipedia structures. The resulting expansion system is called Wiki-MetaSemantik. We tested this system against other online thesauruses and ontology based QE in both individual and meta-search engines setups. Despite that our system has to build a Wikipedia ontology graph in order to do its work, the technique turns out to work very fast (1:281) compared to another ontology QE baseline (Wikipedia Persian ontology QE). It has thus the potential to be utilized online. Furthermore, it shows significant improvement in accuracy. Wiki-MetaSemantik also shows better performance in a meta-search engine (MSE) set up rather than in an individual search engine set up
Highly Relevant Routing Recommendation Systems for Handling Few Data Using MDL Principle and Embedded Relevance Boosting Factors
Full text linkA route recommendation system can provide better recommendation if it also takes collected user reviews into account, e.g. places that generally get positive reviews may be preferred. However, to classify sentiment, many classification algorithms existing today suffer in handling small data items such as short written reviews. In this paper we propose a model for a strongly relevant route recommendation system that is based on an MDL-based (Minimum Description Length) sentiment classification and show that such a system is capable of handling small data items (short user reviews). Another highlight of the model is the inclusion of a set of boosting factors in the relevance calculation to improve the relevance in any recommendation system that implements the model
Centrality dependence of inclusive J/psi production in p-Pb collisions at √S-NN=5.02TeV
Full text linkWe present a measurement of inclusive J/ψ production in p-Pb collisions at sNN−−−√=5.02 TeV as a function of the centrality of the collision, as estimated from the energy deposited in the Zero Degree Calorimeters. The measurement is performed with the ALICE detector down to zero transverse momentum, p T, in the backward (−4.46 < y cms < −2.96) and forward (2.03 < y cms < 3.53) rapidity intervals in the dimuon decay channel and in the mid-rapidity region (−1.37 < y cms < 0.43) in the dielectron decay channel. The backward and forward rapidity intervals correspond to the Pb-going and p-going direction, respectively. The p T-differential J/ψ production cross section at backward and forward rapidity is measured for several centrality classes, together with the corresponding average p T and p T2 values. The nuclear modification factor is presented as a function of centrality for the three rapidity intervals, and as a function of p T for several centrality classes at backward and forward rapidity. At mid- and forward rapidity, the J/ψ yield is suppressed up to 40% compared to that in pp interactions scaled by the number of binary collisions. The degree of suppression increases towards central p-Pb collisions at forward rapidity, and with decreasing p T of the J/ψ. At backward rapidity, the nuclear modification factor is compatible with unity within the total uncertainties, with an increasing trend from peripheral to central p-Pb collisions
Centrality dependence of inclusive J/psi production in p-Pb collisions at √S-NN=5.02TeV
No full textWe present a measurement of inclusive J/ψ production in p-Pb collisions at sNN−−−√=5.02 TeV as a function of the centrality of the collision, as estimated from the energy deposited in the Zero Degree Calorimeters. The measurement is performed with the ALICE detector down to zero transverse momentum, p T, in the backward (−4.46 < y cms < −2.96) and forward (2.03 < y cms < 3.53) rapidity intervals in the dimuon decay channel and in the mid-rapidity region (−1.37 < y cms < 0.43) in the dielectron decay channel. The backward and forward rapidity intervals correspond to the Pb-going and p-going direction, respectively. The p T-differential J/ψ production cross section at backward and forward rapidity is measured for several centrality classes, together with the corresponding average p T and p T2 values. The nuclear modification factor is presented as a function of centrality for the three rapidity intervals, and as a function of p T for several centrality classes at backward and forward rapidity. At mid- and forward rapidity, the J/ψ yield is suppressed up to 40% compared to that in pp interactions scaled by the number of binary collisions. The degree of suppression increases towards central p-Pb collisions at forward rapidity, and with decreasing p T of the J/ψ. At backward rapidity, the nuclear modification factor is compatible with unity within the total uncertainties, with an increasing trend from peripheral to central p-Pb collisions
- …
